Turning Adversaries into Allies: Reversing Typographic Attacks for Multimodal E-Commerce Product Retrieval

TL;DR

This paper introduces a method that enhances multimodal e-commerce product retrieval by rendering textual metadata onto images, improving model robustness against typographic attacks and increasing retrieval accuracy across various datasets and models.

Contribution

It proposes a novel vision-text compression technique that reverses typographic attacks, strengthening image-text alignment for better retrieval performance.

Findings

Consistent improvement in retrieval accuracy across datasets and models.

Effective enhancement for zero-shot multimodal retrieval.

Visually rendering metadata boosts model robustness against attacks.

Abstract

Multimodal product retrieval systems in e-commerce platforms rely on effectively combining visual and textual signals to improve search relevance and user experience. However, vision-language models such as CLIP are vulnerable to typographic attacks, where misleading or irrelevant text embedded in images skews model predictions. In this work, we propose a novel method that reverses the logic of typographic attacks by rendering relevant textual content (e.g., titles, descriptions) directly onto product images to perform vision-text compression, thereby strengthening image-text alignment and boosting multimodal product retrieval performance. We evaluate our method on three vertical-specific e-commerce datasets (sneakers, handbags, and trading cards) using six state-of-the-art vision foundation models. Our experiments demonstrate consistent improvements in unimodal and multimodal retrieval accuracy across categories and model families. Our findings suggest that visually rendering product metadata is a simple yet effective enhancement for zero-shot multimodal retrieval in e-commerce applications.